# Harleen Kaur — enterprise AI engineering

> Open-source work in LLM model routing, agent governance, containment testing, and application evaluation. This portfolio links runnable examples and evidence with explicit limitations.

The featured customer-summary workflow connects AxonLLM, Ostiari, and Escape Lab through a deterministic synthetic fixture. Practical Eval Lab supplies separate application-evaluation examples. The public material supports inspection of engineering work; it does not establish production adoption, business savings, executive scope, or general safety certification.

## Start here
- [Portfolio in Markdown](https://hk-775.github.io/hk-775/index.md): The four projects, their relationship, and runnable entry points.
- [Engineering case study](https://hk-775.github.io/hk-775/case-study.md): The customer-summary boundary, paired results, exact source references, and limitations.
- [Engineering decisions](https://hk-775.github.io/hk-775/decisions.md): Product boundaries, governance choices, and evidence practices.
- [Run the integration](https://hk-775.github.io/hk-775/run-example.md): Locked installation and reproduction commands.
- [Engineering blog](https://hk-775.github.io/hk-775/blog/): Articles on AI workloads, design decisions, and evaluation evidence.
- [Designing an Evidence Trail for Agent Actions](https://hk-775.github.io/hk-775/blog/designing-an-evidence-trail-for-agent-actions.html): I trace one synthetic agent action from its request to its observed effect, then examine what event hashes, snapshots, and replay can establish—and what needs additional verification.
- [When rules beat decision models](https://hk-775.github.io/hk-775/blog/when-rules-beat-decision-models.html): I replaced a label-matching diagnostic with an executable support workflow. On 144 synthetic test episodes, the rules baseline outperformed the tested zero-shot model configurations.

## Evidence and context
- [Paired result data](https://hk-775.github.io/hk-775/evidence/summary.json): One recorded synthetic run per control profile.
- [Combined context download](https://hk-775.github.io/hk-775/agent-context.txt): 7 allowlisted public documents, including blog articles, with source URLs and SHA-256 fingerprints.
- [Blog RSS feed](https://hk-775.github.io/hk-775/blog/feed.xml): Published engineering articles.
- [Source manifest](https://hk-775.github.io/hk-775/discovery.json): Document provenance and project destinations in JSON.
- [Eval Lab agent guide](https://hk-775.github.io/practical-eval-lab/llms.txt): Six evaluation walkthroughs, contracts, and recorded results.

## Optional
- [Coding-agent instructions](https://hk-775.github.io/hk-775/coding-agents.md): Setup, tests, repository map, and evidence-handling conventions.
- [Professional contact](https://www.linkedin.com/in/harleenkaurprofile): Harleen Kaur's LinkedIn profile for AI engineering opportunities and collaboration.
- [GitHub profile](https://github.com/hk-775): Repository ownership and public activity.